1. Clinical and Genomic Studies I have recruited and examined 169 families where at least one member is affected by uveal coloboma. A total of 654 records included patients screened at other sites and patients who were ultimately diagnoses with other developmental diseases. All 169 probands and their first-degree relatives (when available) have complete ophthalmic exams. General physical examinations and targeted systemic testing were performed on probands, as needed. Lymphoblastoid cell lines were established on all participants. We performed targeted sequencing of 196 genes using a custom-capture, high-throughput sequencing platform (Illumina HiSeq2000) in 217 study subjects from 66 families largely with non-syndromic ocular coloboma condition. The 196 known genes on the platform included established coloboma genes, candidate genes from our unbiased screen of laser-capture microdissected (LCM) developing mouse tissue (see below), and genes known to cause coloboma in animal models. We identified and confirmed by Sanger sequencing novel variants in genes previously associated with coloboma, such as RARB and BMP7(missense), TFAP2A (frameshift) and CHD7 (nonsense) in four unrelated families. The novel missense mutation in the retinoid acid receptor (RAR)B gene affected the highly conserved DNA-binding domain and resulted in structural and functional changes within the RARB protein that ultimately affect retinoic acid signaling. Furthermore, we identified a role for RARB and its RXR heterodimeric partners in RA signaling during optic fissure closure in eye development. One of our goals was to test the applicability of custom capture targeted sequencing for the identification of novel variants in uveal coloboma. However, because of high genetic heterogeneity, this methodology had a very low yield suggesting extensive sequencing approaches, such as whole exome or genome sequencing (WES/WGS), are more appropriate for this disease. Families with coloboma are now offered participation in the WES/WGS protocol and so far, we have enrolled 99 coloboma participants. 2. Laboratory Studies A. The RICO mouse model of Coloboma. The RICO (Retinal & Iris COloboma) mouse arose from the random insertion of a transgene (NSE-VEGF) on chromosome 13 in the C57BL/6 background. Both homozygous and heterozygous mutants develop coloboma and the homozygous transgenic mouse is perinatally lethal. We have extensively characterized the genomic organization at the insertion site that includes approximately 30 copies of the transgene, an inversion, three duplications and a deletion in a gene desert. We have detected transient mRNA expression during eye and brain development but detection of the exogenous protein remains challenging. Since the homozygous transgenic mouse is embryonically lethal, we have outcrossed the RICO mouse from the C57BL/6 background to the 129/SvJ. The F1 mouse on 50:50 background survives in the homozygous state and the severity of the coloboma phenotype appears increased. We are now further backcrossing into the 129/SvJ background to F2 and F3 to test the persistence of the phenotype and survival of the homozygous mutants. B. Identification of coloboma candidate genes by molecular characterization of gene expression during optic fissure closure. Our published gene expression study of the margins of the optic fissure (isolated by LCM) in the developing mouse at the time of closing had yielded a number of candidate genes, including the transcriptional regulators Zfp703/Nlz1, Zfp503/Nlz2 and FAT1 protocadherin that we are currently pursuing. The two zinc-finger motif-containing genes, Nlz1 and Nlz2, are important in regulating optic fissure closure in zebrafish. Nlz1 morphant fish display multiple phenotypes such as cystic kidneys and abnormalities in heart development, likely representing a syndromic form of coloboma. We generated mice carrying targeted loxp sites (floxed) to conditionally knockout (KO) Nlz1 and Nlz2. Homozygous floxed Nlz1 mice retaining the Neo cassette were embryonically lethal and the surviving heterozygous showed a syndromic phenotype, including coloboma. Conditional KO mice are currently being generated. Nlz2 KO mice were generated by crossing to beta actin-Cre mice. Nlz2 KO mice were perinatally lethal. Although overall morphologically normal, they exhibited coloboma. Differential expression by RNA-Seq of RPE at E11.5 from Nlz2 KO compared to WT suggested the involvement of signaling pathways and extracellular matrix genes that are currently being explored. Protein expression studies in mouse eye indicated that Nlz2 is expressed transiently during development in the RPE, and in developing and adult amacrine and ganglion cells. We have not yet identified Nlz2 mutations in our coloboma patient DNA samples. A manuscript describing these data is in preparation. FAT protocadherins were among the candidate gene families suggested by our LCM screen. FAT1 is highly expressed during optic fissure closure at the fissure margins, in the periocular mesenchyme and in the optic cup. Fat1 knockout mice and zebrafish embryos homozygous for truncating fat1a mutations exhibit completely penetrant coloboma and recapitulate the most consistent developmental defect observed in a cohort of patients carrying homozygous truncating mutations in FAT1. Additional phenotypes in the patients include facial dysmorphism, blepharoptosis and syndactyly with or without nephropathy, suggestive of a novel syndrome. In our cellular and molecular studies, we first focused on FAT1 role in retinal pigment epithelium (RPE) cells, the primary site of fusion of the optic fissure margins during vertebrate eye development. FAT1 protein was expressed at the earliest cell-cell junctions in human fetal (hf)RPE, placing it exquisitely at the precise cell type and location for facilitating optic fissure fusion. In developing zebrafish eyes, FAT1-positive cell processes emanating from the opposing edges of the fissure margins formed simple appositional-type contacts during the early stages of optic fissure closure. Knockdown of FAT1 resulted in disruption of MENA and ZO1 expression at the earliest cell-cell junctions in hfRPE. Furthermore, CRISPR/Cas9-mediated disruption of MENA or the PDX binding domains of Fat1a were sufficient to cause coloboma in zebrafish. Our data support an essential role for FAT1 in RPE during the earliest cell-cell interactions necessary for completing the developmental program leading to optic fissure closure. C. CRISPR screening for genes associated with optic fissure closure We have developed two experimental pipelines to systematically knock out using CRISPR/Cas9 technology and phenotypically screen in zebrafish all 164 candidate coloboma genes identified from our LCM screen and gene expression study in mouse. The first approach was single gene targeting. To date, we have generated zebrafish founders (F0) carrying CRISPR/Cas9-induced insertion-deletions or large deletions in 69 candidate genes. Of the 55 lines screened, 20 showed eye phenotypes, including 13 lines with coloboma in the first generation (F1). For some of the older lines, we have generated and are analyzing F2 offspring. We are planning to pursue further characterization of the most interesting candidates. Concurrently, we tested a multiplexing CIRSPR/Cas9 targeting approach in which sgRNAs designed to match five genes located on different chromosomes were co-injected into 2-cell stage zebrafish embryos. Although we were successful in generating INDELs in multiple genes at once, none of our multiplex-targeted fish exhibited coloboma. Thus, we abandoned this approach as it ultimately was less efficient than single gene targeting.